Bug#910237: Bug

[Ivo De Decker]

Control: severity -1 serious

Hi,

On Mon, Oct 08, 2018 at 12:41:46PM -0500, Steve Robbins wrote:
> On Monday, October 8, 2018 1:58:13 AM CDT Graham Inggs wrote:
> > Hi Steve / Doug
> > 
> > On Mon, 8 Oct 2018 at 07:27, Steve Robbins <steve at sumost.ca> wrote:
> > > This level seems a bit extreme, to me, considering the guidelines in
> > > https:// www.debian.org/Bugs/Developer
> > 
> > Severity serious is correct.  From the RC policy document [1]:
> > 
> > Packages must autobuild without failure on all architectures on which
> > they are supported.
> 
> Right, and googletest autobuilds while mathicgb does not.  So my reading is 
> that this criteria applies to a bug in mathicgb, not googletest.
>  
> > > Moreover, it's not clear that the bug lies with googletest.
> > 
> > Yes, there is at least an RC bug in googletest or mathicgb, hence Paul
> > wrote:
> > On Wed, 3 Oct 2018 at 20:27, Paul Gevers <elbrus at debian.org> wrote:
> > > Due to the nature of this issue, I filed
> > > this bug report against both packages. Can you please investigate the
> > > situation and reassign the bug to the right package?
> 
> Yes.  Clearly there is a change in the interface that mathicgb is using.  I 
> can't tell whether the change to googletest was incidental or deliberate.  In 
> the former case, there would be a bug in googletest.

As long as this isn't clear, the bug should remain filed against both
packages.

> However, it would not be 
> serious in my reading of the criteria ("is a severe violation of Debian policy 
> (roughly, it violates a must or required directive), or, in the package 
> maintainer's or release manager's opinion, makes the package unsuitable for 
> release").

If it turns out the bug is in googletest, then it is serious, because it
causes other packages to FTBFS. This makes the package unsuitable for release.

If it is not a bug in googletest, the bug should be reassigned to mathicgb. In
that case it is an FTBFS bug, so it should be serious. In this case I'm not
sure if it's reasonable to let googletest migrate without a fix for mathicgb.
In comparable situations, the reverse dependencies are usually given some time
to be fixed.

> > Googletest 1.8.1-1 should not migrate to testing as long as mathicgb
> > FTBFS.  
> 
> This is the assertion that I don't understand.

This package would introduce an FTBFS bug in testing. That's a good reason not
to let it migrate. If it turns out mathicgb is broken, then googletest
shouldn't migrate as long as the broken version of mathicgb is in testing.

Cheers,

Ivo